*** Not Getting the Message on Climate? Attention as a Key Barrier to Mass-Marketing Experimentally-Validated Messages 

*** Nicholas Carnes, Duke University 
*** Geoffrey L. Henderson, University of Michigan

*** Replication Code, version 1.0 ***
*** Updated 28 May 2025



************************************
*** Appendix Table A1 / Figure 1 ***
************************************

[[RESEARCHER: Replace this line with code that opens the file "Study 1 Replication Data.dta" on your machine]]

* Models below appear in the same order as in Table A1
* Please note that the order of categories in Figure 1 is slightly different, to facilitate easier visual interpretation of the figure. 

* DV: Happening

regress dv_when1minus0 i.MAILER
regress dv_when1minus0 any_mailer

* DV: Human-caused

regress dv_cause1minus0 i.MAILER
regress dv_cause1minus0 any_mailer

* DV: Concerning (Six Amer.)

regress dv_sassy1minus0 i.MAILER
regress dv_sassy1minus0 any_mailer

* DV: Concerning (Pew)

regress dv_con1minus0 i.MAILER
regress dv_con1minus0 any_mailer

* DV: Most Important


regress dv_cc1minus0 i.MAILER
regress dv_cc1minus0 any_mailer

* DV: Concerning to Republicans

regress dv_rep1mins0 i.MAILER
regress dv_rep1mins0 any_mailer


************************************
*** Appendix Table A2 / Figure 2 ***
************************************

[[RESEARCHER: Replace this line with code that opens the file "Study 2 Replication Data.dta" on your machine]]

regress happening low medium high
regress humancaused low medium high
regress concern low medium high
regress sixamericas low medium high 


*************************************************************************
*** Appendix Table A3: Regression MOdels for Figure 2 (with controls) ***
*************************************************************************

regress happening low medium high IDEOL_* EDUC_* KNOW_* MEDI_* yearborn i.probiz
regress humancaused low medium high IDEOL_* EDUC_* KNOW_* MEDI_* yearborn i.probiz
regress concern low medium high IDEOL_* EDUC_* KNOW_* MEDI_* yearborn i.probiz
regress sixamericas low medium high  IDEOL_* EDUC_* KNOW_* MEDI_* yearborn i.probiz

**********************************************************
*** Appendix Table A4: CACE Estimates from 2SLS Models ***
**********************************************************

regress treated medhigh if (postcard==0 | medhigh==1)
drop treated_star
predict treated_star, xb

regress happening treated_star if (postcard==0 | medhigh==1)
regress humancaused treated_star if (postcard==0 | medhigh==1)
regress concern treated_star if (postcard==0 | medhigh==1)
regress sixamericas treated_star if (postcard==0 | medhigh==1)

********************************************************************************************************************
*** Table A5: Regression Models for Individual Postcards in Study 2 (Medium- and High Attention Groups Combined) ***
********************************************************************************************************************

gen scientific = regexm(Condition, "Scientific Consensus")
gen publichealth = regexm(Condition, "Public Health")
gen identity = regexm(Condition, "Conservative Identity")
gen adversarial = regexm(Condition, "Team Oriented")

regress happening scientific publichealth identity adversarial if (postcard==0 | medhigh==1)
regress humancaused scientific publichealth identity adversarial if (postcard==0 | medhigh==1)
regress concern scientific publichealth identity adversarial if (postcard==0 | medhigh==1)
regress sixamericas scientific publichealth identity adversarial if (postcard==0 | medhigh==1)

***********************************************************************************
*** Appendix Table A6: Regression Models for Figure 2 (by Primary Voter Status) ***
***********************************************************************************

regress happening low medium high if nonprimary==0
regress happening low medium high if nonprimary==1
regress humancaused low medium high if nonprimary==0
regress humancaused low medium high if nonprimary==1
regress concern low medium high if nonprimary==0
regress concern low medium high if nonprimary==1
regress sixamericas low medium high  if nonprimary==0
regress sixamericas low medium high  if nonprimary==1


***********************************************
*** Table A7: Regression Models for Study 3 ***
***********************************************

[[RESEARCHER: Replace this line with code that opens the file "Study 3 Replication Data.dta" on your machine]]

* DV: Happening	
regress changewhenhappen scratchoff

* DV:Human-caused	
regress changehumancaused scratchoff

* DV: Concerning (Pew)
regress changeSASSY scratchoff

* DV: Concerning (6 Amer.)	
regress changepew scratchoff

* DV: Most Imp. Prob.
regress change_climate scratchoff